Original title: Deep Neural Networks Based Recognition of Plant Diseases by Leaf Image Classification
The researchers train a computer to sort pictures of leaves into healthy leaves or one of 13 disease types. It uses a convolutional neural network, a kind of system that learns visual patterns from images. The model can also distinguish leaves from the surroundings in a picture. In the reported separate-class tests, precisionThe share of the system's predictions for a category that are correct. ranges from 91% to 98%, averaging 96.3%.
Imagine sorting a pile of leaves into healthy ones and those with different signs of disease. It is hard when the leaves are mixed with other things, and some may look alike. In this paper, sorting each leaf matches the system's task of assigning an image to a healthy or diseased category, while separating leaves from their surroundings helps it focus on what it is sorting.
Plant diseases can be hard to spot by eye, and a way to classify leaf images could help people recognize them. The paper reports a model tested on separate disease classes, but the abstract does not show how it performs in everyday farm conditions.
The abstract reports precision from separate-class tests but does not describe performance in real-world farm settings, the size or composition of the image collection, or how well the model transfers to other plants and conditions. It does not establish that the system can diagnose disease from anything beyond leaf images.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Deep Neural Networks Based Recognition of Plant Diseases by Leaf Image Classification — Srdjan Sladojević, Marko Arsenović, Andraš Anderla et al. — Computational Intelligence and Neuroscience — doi:10.1155/2016/3289801
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper.